4 * The contents of this file are subject to the terms of the
5 * Common Development and Distribution License (the "License").
6 * You may not use this file except in compliance with the License.
8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9 * or http://www.opensolaris.org/os/licensing.
10 * See the License for the specific language governing permissions
11 * and limitations under the License.
13 * When distributing Covered Code, include this CDDL HEADER in each
14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15 * If applicable, add the following below this CDDL HEADER, with the
16 * fields enclosed by brackets "[]" replaced with your own identifying
17 * information: Portions Copyright [yyyy] [name of copyright owner]
22 * Copyright (c) 2011, Lawrence Livermore National Security, LLC.
26 #include <sys/zfs_vfsops.h>
27 #include <sys/zfs_vnops.h>
28 #include <sys/zfs_znode.h>
33 zpl_open(struct inode
*ip
, struct file
*filp
)
39 error
= -zfs_open(ip
, filp
->f_mode
, filp
->f_flags
, cr
);
41 ASSERT3S(error
, <=, 0);
46 return generic_file_open(ip
, filp
);
50 zpl_release(struct inode
*ip
, struct file
*filp
)
56 error
= -zfs_close(ip
, filp
->f_flags
, cr
);
58 ASSERT3S(error
, <=, 0);
64 zpl_readdir(struct file
*filp
, void *dirent
, filldir_t filldir
)
66 struct dentry
*dentry
= filp
->f_path
.dentry
;
71 error
= -zfs_readdir(dentry
->d_inode
, dirent
, filldir
,
74 ASSERT3S(error
, <=, 0);
79 ZPL_FSYNC_PROTO(zpl_fsync
, filp
, unused_dentry
, datasync
)
85 error
= -zfs_fsync(filp
->f_path
.dentry
->d_inode
, datasync
, cr
);
87 ASSERT3S(error
, <=, 0);
93 zpl_read_common(struct inode
*ip
, const char *buf
, size_t len
, loff_t pos
,
94 uio_seg_t segment
, int flags
, cred_t
*cr
)
100 iov
.iov_base
= (void *)buf
;
106 uio
.uio_loffset
= pos
;
107 uio
.uio_limit
= MAXOFFSET_T
;
108 uio
.uio_segflg
= segment
;
110 error
= -zfs_read(ip
, &uio
, flags
, cr
);
114 return (len
- uio
.uio_resid
);
118 zpl_read(struct file
*filp
, char __user
*buf
, size_t len
, loff_t
*ppos
)
124 read
= zpl_read_common(filp
->f_mapping
->host
, buf
, len
, *ppos
,
125 UIO_USERSPACE
, filp
->f_flags
, cr
);
136 zpl_write_common(struct inode
*ip
, const char *buf
, size_t len
, loff_t pos
,
137 uio_seg_t segment
, int flags
, cred_t
*cr
)
143 iov
.iov_base
= (void *)buf
;
149 uio
.uio_loffset
= pos
;
150 uio
.uio_limit
= MAXOFFSET_T
;
151 uio
.uio_segflg
= segment
;
153 error
= -zfs_write(ip
, &uio
, flags
, cr
);
157 return (len
- uio
.uio_resid
);
161 zpl_write(struct file
*filp
, const char __user
*buf
, size_t len
, loff_t
*ppos
)
167 wrote
= zpl_write_common(filp
->f_mapping
->host
, buf
, len
, *ppos
,
168 UIO_USERSPACE
, filp
->f_flags
, cr
);
179 * It's worth taking a moment to describe how mmap is implemented
180 * for zfs because it differs considerably from other Linux filesystems.
181 * However, this issue is handled the same way under OpenSolaris.
183 * The issue is that by design zfs bypasses the Linux page cache and
184 * leaves all caching up to the ARC. This has been shown to work
185 * well for the common read(2)/write(2) case. However, mmap(2)
186 * is problem because it relies on being tightly integrated with the
187 * page cache. To handle this we cache mmap'ed files twice, once in
188 * the ARC and a second time in the page cache. The code is careful
189 * to keep both copies synchronized.
191 * When a file with an mmap'ed region is written to using write(2)
192 * both the data in the ARC and existing pages in the page cache
193 * are updated. For a read(2) data will be read first from the page
194 * cache then the ARC if needed. Neither a write(2) or read(2) will
195 * will ever result in new pages being added to the page cache.
197 * New pages are added to the page cache only via .readpage() which
198 * is called when the vfs needs to read a page off disk to back the
199 * virtual memory region. These pages may be modified without
200 * notifying the ARC and will be written out periodically via
201 * .writepage(). This will occur due to either a sync or the usual
202 * page aging behavior. Note because a read(2) of a mmap'ed file
203 * will always check the page cache first even when the ARC is out
204 * of date correct data will still be returned.
206 * While this implementation ensures correct behavior it does have
207 * have some drawbacks. The most obvious of which is that it
208 * increases the required memory footprint when access mmap'ed
209 * files. It also adds additional complexity to the code keeping
210 * both caches synchronized.
212 * Longer term it may be possible to cleanly resolve this wart by
213 * mapping page cache pages directly on to the ARC buffers. The
214 * Linux address space operations are flexible enough to allow
215 * selection of which pages back a particular index. The trick
216 * would be working out the details of which subsystem is in
217 * charge, the ARC, the page cache, or both. It may also prove
218 * helpful to move the ARC buffers to a scatter-gather lists
219 * rather than a vmalloc'ed region.
222 zpl_mmap(struct file
*filp
, struct vm_area_struct
*vma
)
224 znode_t
*zp
= ITOZ(filp
->f_mapping
->host
);
227 error
= generic_file_mmap(filp
, vma
);
231 mutex_enter(&zp
->z_lock
);
233 mutex_exit(&zp
->z_lock
);
239 * Populate a page with data for the Linux page cache. This function is
240 * only used to support mmap(2). There will be an identical copy of the
241 * data in the ARC which is kept up to date via .write() and .writepage().
243 * Current this function relies on zpl_read_common() and the O_DIRECT
244 * flag to read in a page. This works but the more correct way is to
245 * update zfs_fillpage() to be Linux friendly and use that interface.
248 zpl_readpage(struct file
*filp
, struct page
*pp
)
257 ASSERT(PageLocked(pp
));
258 ip
= pp
->mapping
->host
;
259 off
= page_offset(pp
);
260 i_size
= i_size_read(ip
);
261 ASSERT3S(off
, <, i_size
);
264 len
= MIN(PAGE_CACHE_SIZE
, i_size
- off
);
268 /* O_DIRECT is passed to bypass the page cache and avoid deadlock. */
269 wrote
= zpl_read_common(ip
, pb
, len
, off
, UIO_SYSSPACE
, O_DIRECT
, cr
);
273 if (!error
&& (len
< PAGE_CACHE_SIZE
))
274 memset(pb
+ len
, 0, PAGE_CACHE_SIZE
- len
);
281 ClearPageUptodate(pp
);
285 flush_dcache_page(pp
);
294 * Write out dirty pages to the ARC, this function is only required to
295 * support mmap(2). Mapped pages may be dirtied by memory operations
296 * which never call .write(). These dirty pages are kept in sync with
297 * the ARC buffers via this hook.
299 * Currently this function relies on zpl_write_common() and the O_DIRECT
300 * flag to push out the page. This works but the more correct way is
301 * to update zfs_putapage() to be Linux friendly and use that interface.
304 zpl_writepage(struct page
*pp
, struct writeback_control
*wbc
)
313 ASSERT(PageLocked(pp
));
314 ip
= pp
->mapping
->host
;
315 off
= page_offset(pp
);
316 i_size
= i_size_read(ip
);
319 len
= MIN(PAGE_CACHE_SIZE
, i_size
- off
);
323 /* O_DIRECT is passed to bypass the page cache and avoid deadlock. */
324 read
= zpl_write_common(ip
, pb
, len
, off
, UIO_SYSSPACE
, O_DIRECT
, cr
);
333 ClearPageUptodate(pp
);
344 const struct address_space_operations zpl_address_space_operations
= {
345 .readpage
= zpl_readpage
,
346 .writepage
= zpl_writepage
,
349 const struct file_operations zpl_file_operations
= {
351 .release
= zpl_release
,
352 .llseek
= generic_file_llseek
,
355 .readdir
= zpl_readdir
,
360 const struct file_operations zpl_dir_file_operations
= {
361 .llseek
= generic_file_llseek
,
362 .read
= generic_read_dir
,
363 .readdir
= zpl_readdir
,